Hier finden Sie wissenschaftliche Publikationen aus den Fraunhofer-Instituten.

Towards automated aim determination of a social media group: A survey as baseline

: Wogatzke, Christian
: Khondoker, Rahamatullah

Volltext urn:nbn:de:0011-n-4237100 (4.7 MByte PDF)
MD5 Fingerprint: 1b34ae8b229eb9245df7e79232ff01a7
Erstellt am: 7.12.2016

Darmstadt, 2016, 18 S.
Darmstadt, TU, Studienarbeit, 2016
Studienarbeit, Elektronische Publikation
Fraunhofer SIT ()
aim determination; sentiment analysis; opinion mining; text classification; topic classification; Social Media; personal trait inference; ontology development

The ubiquity of social media offers a wide range of research topics not least concerning information retrieval tasks. While numerous publications have adressed tasks such as opinion mining, sentiment analysis or domain-specific feature extraction in order to gain information about real world events, sentiments of participants in extremist forums or to improve information filtering, less attention has been payed on opinions, sentiments or features of an entire social media group. Moreover, the determination of the aim of such a group could leverage the exposure of extremist or riot groups. For this purpose a method for automated aim determination of a social media group is presented. As a first step a research on related building blocks is conducted. On this basis a literature survey on valuable publications is presented. Supplementary to common formats of survey papers a structured table is provided, which gives an overview on the examined resources. For each resource listed in the table a corresponding flow text outlines its specific content. This survey then functions as a baseline for the proposed aim determination method. Furthermore, emerging method requirements are depicted and a framework architecture is suggested. The method is evaluated by the means of existing approaches of the requirements, which emerge within the scope of the method. Finally, a method is established as a profound basis for further research.